AKR1B10 Serves as a Specific Biomarker for the Identification of Hepatocellular Ballooning Degeneration in MASH Patients

BACKGROUND: Hepatocellular ballooning degeneration distinguishes metabolic dysfunction-associated steatohepatitis (MASH) from metabolic dysfunction-associated steatotic liver (MASL), yet specific biomarkers are lacking. METHODS: The spatial transcriptomes of liver sections from human MASL and MASH patients and single-cell data were used to examine hepatocyte ballooning. AKR1B10 and CK8/18 expression in 20 human liver samples were analysed by H&E staining and immunofluorescence. Serum AKR1B10 levels were quantified in 284 patients with various liver aetiologies to assess its differential expression across liver disease types and its diagnostic performance. RESULTS: Spatial and single-cell transcriptomic analyses revealed significant AKR1B10 enrichment in the ballooned hepatocytes. The proportions of T cells and fibroblasts were significantly greater in ballooning regions than in steatotic regions. Immunostaining confirmed the localization of AKR1B10 in ballooning regions, along with the cytoplasm and cell membrane. This expression pattern was consistent with areas of cytokeratin CK8/18 loss; moreover, their combined application improved the ballooning diagnostic rate and accuracy. Serum AKR1B10 levels were markedly higher in patients with MASH than in individuals with other liver diseases (p < 0.01) and correlated with ballooning severity (R = 0.624). Ordinal logistic regression analysis indicated that the serum AKR1B10 level was strongly positively correlated with liver tissue ballooning severity (OR = 1.541 per unit, p = 0.003). The constructed model demonstrated significant discriminatory ability for identifying MASH patients with ballooning (AUC = 0.863). CONCLUSIONS: AKR1B10 is a potential diagnostic and therapeutic biomarker for hepatocyte ballooning in MASH patients with strong spatial specificity and clinical value for noninvasive diagnosis, disease stratification, and treatment monitoring.